Temporal Action Detection on ActivityNet 1.3 (test)
38.5Average mAPDyFADet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DyFADetFeatures=R(2+1)D2024.07 | 38.5 | 58.1 | 39.6 | 8.4 | |
| TriDetFeatures=R(2+1)D2024.07 | 36.8 | 54.7 | 38 | 8.4 | |
| ActionFormerFeatures=R(2+1)D2024.07 | 36.6 | 54.7 | 37.8 | 8.4 | |
| TAGSType=Proposal Free, Backbone=I3D2022.07 | 36.5 | 56.3 | 36.8 | 9.6 | |
| BMNclassifier=[36]2019.07 | 36.42 | — | — | — | |
| Dr2NetDownstream training=End-to-end, Memory=24.1GB2024.01 | 36.3 | — | — | — | |
| ConventionalDownstream training=End-to-end, Memory=44.7GB2024.01 | 36.2 | — | — | — | |
| ASLFeatures=I3D2024.07 | 36.2 | 54.1 | 37.4 | 8 | |
| TAGSType=Proposal Free, Backbone=TS2022.07 | 35.9 | 53.7 | 36.1 | 9.5 | |
| GTAD+CSAType=Actioness, Backbone=TS2022.07 | 35.7 | 51.8 | 36.8 | 8.7 | |
| TALLFormerFeatures=Swin2024.07 | 35.6 | 54.1 | 36.2 | 7.9 | |
| TCANetType=Actioness, Backbone=TS2022.07 | 35.5 | 52.2 | 36.7 | 6.8 | |
| TCANetFeatures=TSN2024.07 | 35.5 | 52.3 | 36.7 | 6.9 | |
| ReversibleDownstream training=Hard finetune, Memory=24.1GB2024.01 | 35.4 | — | — | — | |
| PBR-NetType=Anchor, Backbone=I3D2022.07 | 35 | 53.9 | 34.9 | 8.9 | |
| VSGNType=Anchor, Backbone=I3D2022.07 | 35 | 52.3 | 36 | 8.3 | |
| PBRNetFeatures=I3D2024.07 | 35 | 54 | 35 | 9 | |
| TadMLSingle-stage=true2022.06 | 34.94 | 53.15 | 35.75 | 7.47 | |
| BSN++Type=Actioness, Backbone=TS2022.07 | 34.8 | 51.2 | 35.7 | 8.3 | |
| VSGNFeatures=I3D2024.07 | 34.7 | 52.3 | 35.2 | 8.3 | |
| TadTRFeatures=TSN2024.07 | 34.6 | 51.3 | 35 | 9.5 | |
| TadTR+BMNSingle-stage=false2022.06 | 34.55 | 50.51 | 35.35 | 8.18 | |
| AFSDSingle-stage=true2022.06 | 34.4 | 52.4 | 35.3 | 6.5 | |
| ConventionalDownstream training=Frozen backbone, Memory=12.2GB2024.01 | 34.4 | — | — | — | |
| GTANType=Anchor, Backbone=P3D2022.07 | 34.3 | 52.6 | 34.1 | 8.9 | |
| BC-GNNType=Actioness, Backbone=TS2022.07 | 34.3 | 50.6 | 34.8 | 9.4 | |
| AFSDFeatures=I3D2024.07 | 34.3 | 52.4 | 35.2 | 6.5 | |
| ContextLocType=Actioness, Backbone=I3D2022.07 | 34.2 | 56 | 35.2 | 3.5 | |
| G-TADType=Actioness, Backbone=TS2022.07 | 34.1 | 50.4 | 34.6 | 9 | |
| G-TADFeatures=TSN2024.07 | 34.1 | 50.4 | 34.6 | 9 | |
| G-TADSingle-stage=true2022.06 | 34.09 | 50.36 | 34.6 | 9.02 | |
| MUSESType=Anchor, Backbone=I3D2022.07 | 34 | 50 | 34.9 | 6.5 | |
| BMNType=Actioness, Backbone=TS2022.07 | 33.9 | 50.1 | 34.8 | 8.3 | |
| BMNFeatures=TSN2024.07 | 33.9 | 50.1 | 34.8 | 8.3 | |
| BMNSingle-stage=false2022.06 | 33.85 | 50.07 | 34.78 | 8.29 | |
| BSNclassifier=[36]2019.07 | 32.87 | — | — | — | |
| BSNClassifier=[46]2018.06 | 32.84 | — | — | — | |
| ReActFeatures=TSN2024.07 | 32.6 | 49.6 | 33 | 8.6 | |
| TadTRSingle-stage=true2022.06 | 32.27 | 49.08 | 32.58 | 8.49 | |
| Lin et al.2018.06 | 32.26 | — | — | — | |
| Lin et al.classifier=[36]2019.07 | 32.26 | — | — | — | |
| RTD-NetType=Actioness, Backbone=I3D2022.07 | 30.8 | 47.2 | 30.7 | 8.6 | |
| RTD-NetFeatures=I3D2024.07 | 30.8 | 47.2 | 30.7 | 8.6 | |
| ContextlocSingle-stage=false2022.06 | 30.59 | 51.24 | 31.4 | 2.83 | |
| BU-TALType=Actioness, Backbone=I3D2022.07 | 30.1 | 43.5 | 33.9 | 9.2 | |
| BSNSingle-stage=false2022.06 | 30.03 | 56.45 | 29.96 | 8.02 | |
| ReversibleDownstream training=From scratch, Memory=24.1GB2024.01 | 28.3 | — | — | — | |
| Zhao et al.Supervision=Fully supervised2017.12 | 28.28 | — | — | — | |
| SSN2018.06 | 28.28 | — | — | — | |
| SSNclassifier=[36]2019.07 | 28.28 | — | — | — | |
| SSNSingle-stage=true2022.06 | 28.28 | 43.26 | 28.7 | 5.63 | |
| SSNmode=single model2017.04 | 28.28 | 43.26 | 28.7 | 5.63 | |
| A2NetType=Mixed, Backbone=I3D2022.07 | 27.8 | 43.6 | 28.7 | 3.7 | |
| A2NetFeatures=I3D2024.07 | 27.8 | 43.6 | 28.7 | 3.7 | |
| A2NetSingle-stage=true2022.06 | 27.75 | 43.55 | 28.69 | 3.7 | |
| P-GCNSingle-stage=false2022.06 | 26.99 | 42.9 | 28.14 | 2.47 | |
| TAGBackbone Architecture=Inception V32017.03 | 26.05 | 40.689 | 26.017 | 6.667 | |
| Xiong et al.Supervision=Fully supervised, ActivityNet Challenge submission=true2017.12 | 26.05 | — | — | — | |
| Temporal Context Networktop 20 proposals=true2017.08 | 23.58 | 37.49 | 23.47 | 4.47 | |
| TCN2018.06 | 23.58 | — | — | — | |
| CDC2018.06 | 22.9 | — | — | — | |
| CDCclassifier=[36]2019.07 | 22.9 | — | — | — | |
| TAL-NetSingle-stage=false2022.06 | 20.22 | 38.23 | 18.3 | 1.3 | |
| TADType=Anchor, Backbone=I3D2022.07 | 20.2 | 38.2 | 18.3 | 1.3 | |
| STPNSupervision=Weakly supervised2017.12 | 20.07 | — | — | — | |
| SCC2018.06 | 19.3 | — | — | — | |
| Singh et. al. [28]2017.03 | 17.83 | 36.398 | 11.05 | 0.14 | |
| Oxfordtop 20 proposals=true2017.08 | 17.83 | 36.4 | 11.05 | 0.14 | |
| Singh & CuzzolinSupervision=Fully supervised, ActivityNet Challenge submission=true2017.12 | 17.83 | — | — | — | |
| Singh et. al. [42]mode=single model2017.04 | 17.83 | 36.4 | 11.05 | 0.14 | |
| Singh et. al. [27]2017.03 | 17.68 | 28.667 | 17.78 | 2.88 | |
| UMDtop 20 proposals=true2017.08 | 17.68 | 28.67 | 17.78 | 2.88 | |
| Singh et al.Supervision=Fully supervised2017.12 | 17.68 | — | — | — | |
| Singh et. al. [41]mode=single model2017.04 | 17.68 | 28.67 | 17.78 | 2.88 | |
| UPCtop 20 proposals=true2017.08 | 14.81 | 22.37 | 14.88 | 4.45 | |
| Wang et. al.2017.03 | 14.62 | 42.478 | 2.88 | 0.06 | |
| QCIStop 20 proposals=true2017.08 | 14.62 | 42.48 | 2.88 | 0.06 | |
| Wang & TaoSupervision=Fully supervised, ActivityNet Challenge submission=true2017.12 | 14.62 | — | — | — | |
| Wang et al.2018.06 | 14.62 | — | — | — | |
| Wang et. al. [54]mode=single model2017.04 | 14.62 | 42.48 | 2.88 | 0.06 | |
| B. Singh et. al.train data=train+val, Feature Type=Handcrafted2019.06 | — | 28.8 | — | — | |
| Dai et. al.train data=train, Feature Type=Handcrafted2019.06 | — | 37.5 | — | — | |
| G. Singh et. al.train data=train, Feature Type=Handcrafted2019.06 | — | 36.4 | — | — | |
| R-C3DType=Anchor, Backbone=C3D2022.07 | — | 26.8 | — | — | |
| R-C3DSingle-stage=false2022.06 | — | 26.8 | — | — | |
| Single-stream R-C3Dtrain data=train, Feature Type=Deep2019.06 | — | 26.8 | — | — | |
| Single-stream R-C3Dtrain data=train+val, Feature Type=Deep2019.06 | — | 28.4 | — | — | |
| UPCtrain data=train, Feature Type=Deep2019.06 | — | 22.3 | — | — | |
| Zhao et. al.train data=train, Feature Type=Deep2019.06 | — | 43.3 | — | — |